Resort Real Estate: An Economic Analysis of Second Home Pricing Behavior in Park City, Utah by Brady W. Larsen B.A., Finance, 2002 B.A., Information Systems, 2002 B.A., Spanish, 2002 University of Utah Submitted to the Program in Real Estate Development in Conjunction with the Center for Real Estate in Partial Fulfillment of the Requirements for the Degree of Master of Science in Real Estate Development at the Massachusetts Institute of Technology September, 2010 ©2010 Brady W. Larsen All rights reserved The author hereby grants to MIT permission to reproduce and to distribute publicly paper and electronic copies of this thesis document in whole or in part in any medium now known or hereafter created. Signature of Author_____________________________________________________________________________ Center for Real Estate August 5, 2010 Certified by___________________________________________________________________________________ William C. Wheaton Professor of Economics Thesis Supervisor Accepted by________________________________________________________________________________ David M. Geltner Chairman, Interdepartmental Degree Program in Real Estate Development 1 Resort Real Estate: An Economic Analysis of Second Home Pricing Behavior in Park City, Utah by Brady W. Larsen Submitted to the Program in Real Estate Development in Conjunction with the Center for Real Estate on August 6, 2010 in Partial Fulfillment of the Requirements for the Degree of Master of Science in Real Estate Development ABSTRACT The purpose of this research project is to examine the market pricing behavior of vacation homes in resort property markets. To accomplish this a price index is constructed to track real price fluctuations from 1981 to 2010 for the 3 localized ski resort markets in Park City, Utah. The resulting price indices reveal a history of cyclical price movements, and surprising long-term real price depreciation of 12% to 25% between 1981 and 2010. To determine the causes of the cyclical movements in the price indices, time series analysis is performed, and a model created to predict market behaviors based on past levels of price, construction, and skier days. The results of this exercise reveal that the number of annual skier days in the area is an effective representative of demand for housing, and that the local ski business has a considerable effect on real estate prices. Additionally, it is revealed that Park City’s ski business is largely affected by national economic conditions, more so than by both regional economical conditions and local snowfall. The analysis concludes that despite the thirty year decline in real prices, the Park City resort market behaves as a well functioning, healthy market. The model indicates that while increases in prices do stimulate new construction, the growth in the total number of dwelling units reveals a relatively inelastic supply market. This suggests that any growth in demand should be accompanied with long-term price appreciation. Market forecasts based on various demand scenarios indicate that except in the most pessimistic cases, prices in Park City should experience healthy appreciation in the near to mid future. It is believed that these findings can be applicable to various resort markets. Thesis Supervisor: William C. Wheaton Title: Professor of Economics 2 Acknowledgements I would like to express gratitude to my thesis supervisor and economics professor William Wheaton for making this project possible with his clear and patient guidance throughout the thesis process. Thanks are also owed to the Park City Board of Realtors for providing the sales transaction data needed to complete this research project. I would also like to thank the faculty and staff of the Center for Real Estate for their contributions to a fantastic educational experience. Finally, I would like to thank my parents, my uncle John Williams, and the rest of my family for their support throughout the academic year. 3 Table of Contents ABSTRACT ................................................................................................................................... 2 Acknowledgements ....................................................................................................................... 3 Table of Contents .......................................................................................................................... 4 1.0 Introduction ............................................................................................................................. 5 1.1 Literature Review ................................................................................................................. 7 2.0 Background ............................................................................................................................. 9 2.1 Park City, Utah ..................................................................................................................... 9 2.1.1 Resorts.......................................................................................................................... 10 3.0 Real Estate Data .................................................................................................................... 12 3.1 Supply ................................................................................................................................. 12 3.1 Price Index .......................................................................................................................... 13 3.2.1 Price Data Collection ................................................................................................... 14 3.2.2 Index Construction ....................................................................................................... 16 3.3 Index Analysis .................................................................................................................... 19 3.3.1 Comparison of Park City to Deer Valley ..................................................................... 21 3.4 Conclusion .......................................................................................................................... 23 4.0 Time Series Analysis ............................................................................................................. 24 4.1 Park City Skier Demand ..................................................................................................... 24 4.1.1 National Economic Data Series ................................................................................... 26 4.1.2 Annual Snowfall .......................................................................................................... 28 4.1.3 Skier Visit Equation ..................................................................................................... 29 4.2 Supply ................................................................................................................................. 30 4.3 Price .................................................................................................................................... 32 5.0 Forecasting Model ................................................................................................................. 35 5.1 Base Forecast ...................................................................................................................... 36 5.2 The Reaction of Forecast to Temporary Shocks ................................................................. 38 5.3 The Reaction of Forecast to Permanent Shocks ................................................................. 41 5.4 Alternative Long-Range Forecasts ..................................................................................... 44 5.4 Forecast Conclusion ............................................................................................................ 47 6.0 Conclusion ............................................................................................................................. 48 Appendices ................................................................................................................................... 51 Appendix 1 – Data ................................................................................................................ 51 Appendix 2 - Regression Results .......................................................................................... 54 Bibliography ................................................................................................................................ 64 4 1.0 Introduction Over the past decades second home development has become more and more prevalent and a strong economic force. Investors have increasingly been purchasing second homes in recreational and resort settings located adjacent to oceans, golf courses, lakes, and mountain resorts. The U.S. Census Bureau estimates that 7.9 million vacation homes exist in the United States today, compared to approximately 75 million owner-occupied homes. According to the National Association of Realtors’ (NAR) 2009 Investment and Vacation Home Buyers Survey the number of vacation homes sold in 2009 increased 7.9% to 553,000, from 513,000 in 2008 - 10% of the overall residential market share1. The increase suggests that buyers are starting to take advantage of bargain prices resulting from the recent economic downturn. The majority of the survey participants indicate that the primary purpose of their new vacation home is to function as a family and recreational retreat. However, 29% of the participants state that portfolio diversification is one of the most important motivators for their purchase. While it is understood that vacation homes can provide an annual yield – whether it be a utility or a rental yield – it is questionable whether or not they can be expected to provide long term appreciation. While the cyclical movements of primary residential markets and commercial property markets have been well researched, there have been few publications that have specifically studied markets for second homes. The objective of this paper is to examine the investment performance and economic behavior of vacation homes in the destination ski resort market of Park City, Utah. Park City is a 4-season resort community, and the home of three destination ski resorts: Park City Mountain Resort, Deer Valley, and The Canyons (located just outside city limits). To complete this study historical residential sales data was collected for sales transactions from 1981 to 2010 for condominiums located near the base of each of the Park City ski resorts. With this data a property price index is constructed for each of the three resorts, to track prices as a function of time from 1981 to 2010. The indices are created by applying multiple regression analysis to the sales data to control for the variable attributes that contribute to the price of 1 National Association of Realtors. Second Homes: Talking Points. 10 March 2010. 6 July 2010 <http://www.realtor.org/press_room_secured/public_affairs/tpsecondhomes>. 5 housing.1 The three price series all reflect similar fluctuation patterns over the index period, and they appear to be very recessionary, reacting largely to the growth of the national economy. Over the 29 year period nominal prices show a moderate overall increase of approximately 100%, while real prices have failed to keep pace with inflation, reflecting a decrease of approximately 18%. It should be noted, however, that after a steep decline between 1981 and 1988 prices trended up considerably until peaking in 2007 before the recent downturn. A comparative study between the three indices is performed, and it is interesting to observe that Deer Valley, considered the more luxurious of the resorts with larger, more expensive units, appreciated less throughout the years of substantial growth, but also appears to have started to recover the soonest. The price series fluctuations for Park City and Deer Valley are next examined using traditional econometrics. External variables such as skier visits (a measure of demand), construction permits (a measure of change in supply), interest rates, regional and national income levels, unemployment levels, job growth, and natural snowfall are gathered to explore the causes of the price fluctuations by way of multiple regression analysis. The price index and variables are used to construct a time series model and a series of equations is assembled as a conditional econometric forecasting model. The series of equations are used to predict skier days, real estate prices, and construction permits. The model reveals that while snowfall does have an effect on the number of annual ski days, the region’s ski business is influenced more by long term economic growth, particularly at the national level, which can be explained by the area’s character as a national ski destination. The study also confirms that real estate price appreciation in the area can largely be explained by the area’s ski business (a measure of demand) as compared to the number of dwelling units in the market. The study concludes that Park City’s supply of residential units is relatively inelastic, such that new supply reacts appropriately to fluctuations in price, indicating that the market is essentially healthy and well behaved. 1 Miller, Norman G. "Residential Property Hedonic Pricing Models: A Review." Research in Real Estate, Vol. 2. JAI Press Inc., 1982. 31-56. 6 To further support the research findings a 15-year conditional forecast model is created to examine the response of skier days, price, and new construction to different economic scenarios: realistic, pessimistic, and optimistic. The model observes impulse responses to exogenous demand shocks that are caused by increases in annual snowfall and national disposable income levels. The market behaves appropriately in all tested scenarios. In response to a forecast of average snowfall and moderate income growth the model predicts a steady increase for both price and stock. In the optimistic scenarios with multiple years of near record snowfall and sustained income growth real estate prices show a dramatic increase, and the new supply market responds with a boom in construction. Even in the most pessimistic scenarios, with snowfall decreasing permanently to near record lows and curtailed economic growth, prices react by dropping considerably, but construction appropriately drops nearly 55% over a 5 year period and prices start to recover in year six. In spite of the 12% decrease in real prices since 1981 which might suggest the contrary, the study results indicate that the real estate market of Park City Utah is a healthy, well behaving market. 1.1 Literature Review The 2005 Journal of Real Estate Research contained a study similar to this paper that examined the New England Ski Market1. In this study Wheaton discovers that real prices of real estate at Loon Mountain Ski Resort depreciated by approximately 40% over a period of 25 years. A similar time series and conditional forecasting model is created which indicates that the New England Ski Market, represented by the number of annual skier days in the region, is largely affected by natural snowfall, more-so than by the region’s long term economic growth or business cycle. The study also indicates that price appreciation at Loon Mountain can be explained closely by the regional ski business in comparison to the stock of units. The examination of the impulse responses in this study revealed that the new supply market at this resort responded so elastically to any movement in price that appreciation would be non-existent due to overbuilding. In nearly all scenarios any positive demand shock would result in a 1 Wheaton, W. C., “Resort Real Estate: Does Supply Prevent Appreciation?” Journal of Real Estate Research ,Vol 27, 2005. 7 building boom, and real prices would eventually fall below the pre-shock levels. Wheaton concludes that investment in the New England Ski Market would not likely produce any real appreciation. In 2008 the MIT Center for Real Estate released a thesis authored by Sean Lee which conducts a similar research study to this and the one authored by Wheaton. Lee creates a price index for properties near Heavenly Ski Resort in the Lake Tahoe, California market for the years 1998 20001. The results of the Tahoe study are drastically different from those of the New England market. Real housing prices in Tahoe remained essentially flat between 1988 and 1998 but then increased nearly 300% until the peak in 2006, before falling 20% through 2008. In contrast to the market in New England, the study determines that the supply market in Tahoe is quite constrained due to its age, size, and stringent building regulations, which seriously impede new development. High demand also plays a role as Tahoe is a true four-season resort that experiences high year-round traffic due to its proximity to the Northern California population, the lake and other summer amenities, as well as the Nevada casinos. The ski business in the Tahoe market is highly affected by both snowfall and regional and national economics as it gets weekend business from all over Northern California, but also serves as a destination resort nationally. As most destination travelers plan their ski vacations long before the snow season begins, their business is less dependent on the current year’s snowpack, and more reliant on the economic growth of the previous year. The papers completed by Wheaton and Lee examine different markets of two very different resorts, and indicate completely unique results. This paper examines the market of Park City, Utah, chosen in part because it also is different from the markets previously studied. The Park City market falls somewhere in the middle of the spectrum between these other two resorts, and it contains many characteristics that might be more typical of a destination ski resort. It is hoped that this study will be able to provide insight into the determinants of price appreciation and cycles in the resort/vacation home industry. 1 Lee, Sean. "Second Home Real Estate Market: Economic Analysis of Residential Pricing Behavior Near Heavenly Ski Resort, CA." 2008. 8 2.0 Background 2.1 Park City, Utah The state of Utah boasts the slogan “Greatest Snow on Earth” and is the home of thirteen ski resorts, eleven of which are located within a one-hour drive of Salt Lake’s International airport, and seven within a 45 minute drive. While most of these resorts have a large number of lifts and extensive trail networks, Park City, Utah is the area that has been most developed into a resort destination with extensive condominium and lodging development, and a vibrant mountain town with restaurants and nightlife. Park City is the home of three world-class ski and summer resorts: Park City Mountain Resort, Deer Valley, and the Canyons (just outside of city limits). The city lies east approximately 36 miles from the Salt Lake International Airport, 32 miles from downtown Salt Lake City, and can be reached with an estimated drive time of 40 minutes via Interstate 80. Park City was first settled in the late 1960’s as a silver-mining town and was incorporated as a city in 1884. The town evolved into a thriving “boom” town and in its heyday at the turn of the century reached a remarkable population of 10,000 and was home to the Silver King Coalition mine, the country’s richest silver mine. With the decline of the mining industry, the population slowly diminished to 1,150 in 1951 and Park City started to decay into a decrepit “ghost” town. However, in 1960 United Park City Mines was looking to diversify, and in 1963 Park City was approved for a federal loan from the Area Redevelopment Agency to open Treasure Mountain (Park City Mountain Resort) on part of a parcel of mining land. The resort opened in 1963 with a gondola, a chairlift, and 2 J-bars, along with a 9-hole golf course, and had 50,000 skier visits its first season1. A sister ski resort named Park City West (The Canyons) was opened 4 miles west of Park City in 1968, and the Deer Valley Resort followed in 19812. The opening of Park City Mountain Resort triggered the evolution of Park City from a decaying mining town to a thriving resort town. The resorts and town have continued to expand steadily up until 2002 when Park 1 Park City Chamber and Visitor's Bureau. "Economic and Relocation Package - Park City History." 2010. ParkCityInfo.com. 5 June 2010 2 Deer Valley actually reopened a small resort called snow park that had operated on and off until it was permanently closed in 1968. 9 City was put into the national spotlight as host of many of the alpine events during the 2002 Salt Lake Olympic Winter Games. Today Park City reports an estimated population of 8,066 residents, as well as a lodging capacity of 23,3071. Tourism is the primary economic driver of the area, as Park City houses approximately 600,000 tourists per year, and receives approximately 3,000,000 visitors per year2. It is also estimated that 60% of the 8,000 (approximate) dwelling units in Park City proper function as second homes. 2.1.1 Resorts Park City Mountain Resort, located just a few blocks from Main Street in downtown Park City, is currently owned by Powdr Corporation, one of the largest ski resort operators in North America. The resort was opened in 1963 with the name of Treasure Mountain by United Park City Mining Co. with one Gondola, a chair lift, and two J-bars. Today the resort consists of 16 chairlifts, 3300 skiable acres, and 3100 vertical feet. The terrain provides skiing for all levels of skiing and snowboarding, including terrain parks to help attract the snowboard population which has grown considerably over recent decades. Park City has long been marketed as one of the higher end destination resorts in the Rocky Mountains. It has been a perennial host of the World Cup since 1985, and hosted 4 different events during the 2002 Winter Olympics. Park City also provides summer recreational opportunities with a concrete sled track called the Alpine Slide, a zip line ride, children carnival rides, miniature golf, as well as lift served mountain biking and hiking. Park City has been voted one of the top 5 resorts in North America in Ski Magazine multiple times, including the most recent poll. Deer Valley is located approximately 1.5 miles east of Park City. The resort opened in 1981 on the former site of a small ski area entitled Snow Park Ski Area that operated on and off between 1946 and 1968 and consisted of just a couple of ski lifts constructed from lodge-pole pines taken directly from the site. Deer Valley is smaller than the other two Park City resorts with 2,026 1 Park City Municipality. "Park City: Quick Facts." 1 1 2010. ParkCity.org. 6 6 2010 <http://www.parkcity.org/index.aspx?page=279>. 2 Park City Chamber and Visitor's Bureau. "Economic and Relocation Package - Tourism." 2010. ParkCityInfo.com. 5 June 2010 10 skiable acres. To compete with nearby resorts Deer Valley has marketed itself as an exclusive high-end resort, catering to a higher-end clientele with amenities such as free ski valets and parking shuttles, fine dining and shopping, more frequent grooming of slopes, and limited access to avoid overcrowding. It is one of only three resorts remaining in North America that does not allow snowboarders. Deer Valley was also host of four different Olympic events during the 2002 games and hosts international freestyle ski events every year. The resort has been named #1 ski resort in North America by Ski Magazine four times in the last eight years, including the three most recent polls. The Canyons Ski Resort opened in 1968 with the name of Park City West as it was located just 4 miles west of its sister resort, Park City Mountain Resort. Its name was changed shortly thereafter to Park West, and then again to Wolf Mountain in 1995. After being purchased in 1997 by American Skiing Company the resort was renamed The Canyons and underwent the start of a $500 million expansion plan that would increase the skiable acres of the resort from 1400 to 3700 by 2007, making it the largest resort in Utah, and one of the 5 largest in the United States. The expansion included major amenity improvements including new lodges, condominiums, and a recently constructed Waldorf Astoria hotel. American Skiing Company was recently dissolved, and the resort was purchased in 2008 by Talisker, a Toronto based real estate development firm. The Canyons is located outside of the Park City municipal boundaries along Highway 224 which connects Interstate 80 to Park City, in an unincorporated area known as South Snyderville Basin. 11 3.0 Real Estate Data 3.1 Supply To appropriately study market pricing behaviors over a specified time period it is necessary to measure the change in supply over that period. In the real estate market, the supply variable is represented by stock, defined as the number of dwelling units located within that market. The change in supply is represented by the amount of new construction within the same market. For this study, the new construction data was provided by the Park City building department, which had tracked the number of residential building permits issued annually within the Park City municipal boundaries from 1980 to 2009. The annual change in supply is therefore calculated simply by using the number of existing dwelling units in Park City as reported in the most recent U.S. census of 2000, and increasing/decreasing that number by the number of new housing construction permits each year. Figure 1 illustrates new construction and total housing supply between 1980 and 2010. A table listing annual housing permits and stock can be found in the Appendix. Housing Supply, Park City New Units 600 10,000 8,000 500 400 6,000 300 4,000 200 Total Units 700 2,000 100 0 - Stock Permits Figure 1 – New Construction & Housing Supply, Park City, Utah - 1980-2010 The number of housing units in Park City encompasses both the Park City and Deer Valley markets as both resorts lie within city boundaries, just over one mile apart. While the available data does not allow differentiation in supply between the two resorts, the stock / permit series are 12 considered good indications of the change in supply for the overall Park City resort market. A follow up study breaking down the overall supply market into submarkets could be interesting.1 3.1 Price Index It can be difficult to track true market-wide price appreciation for housing due to the heterogeneous nature of the housing market. The purchase price of a home can be viewed as the combined value of the multiple attributes that each contribute to the value of that home. Home values are therefore difficult to predict, and to compare apple to apples, due to the fact no two houses are the same. There are many different variables that contribute to the value of a home, including, but not limited to: square footage, number of bedrooms and bathrooms, lot size, age, quality, location, views, and layout. The amount that each individual characteristic adds to the value of a house in a particular market is difficult to discern by mere observation, but can be measured by estimating what is called an hedonic price equation2. An hedonic price equation is an econometric tool that is derived by using multiple regression analysis against a series of data to determine the effect that each observable independent variable has on price, such that price is a function of the observable values of each of its individual attributes, as follows3: Price = α + β1X1 + β2X2 + β3X3 + β4X4 + ….. + βnXn (or) Price = αX1 β1X2 β2X3 β3 X4 β4 ….. + Xnβn (Eq.1.1 – Linear) (Eq.1.2 – Exponential) α - Intercept. (constant affected by ind variables to predict price) X - Independent variable (observed value) β - Coefficient (measure of effect that X has on α) Equation 1 – Hedonic Price Equation 1 Park City housing supply numbers do not represent supply for The Canyons’ housing market as The Canyons is located outside of the Park City municipal boundaries. Building Permits for The Canyons and its surrounding area are issued by Summit County, which has only tracked permits issued annually across the entire county, an area deemed too broad to be effective for this study. 2 Miller, Norman G. "Residential Property Hedonic Pricing Models: A Review." Research in Real Estate, Vol. 2. JAI Press Inc., 1982. 31-56. 3 DiPasquale, Denise and William C. Wheaton. Urban Economics and Real Estate Markets. Prentice-Hall, Inc. , 1996. 13 Hedonic regression analysis is a method commonly used to examine how consumers in a market value certain attributes, and can be beneficial in the process of both appraising existing real estate, and deciding if, what, and where a real estate asset should be built. In a similar fashion an hedonic price equation can also be used to track true changes in price over a period of time. An effective housing price equation has broken down the values of a house into the increments of each of its individual attributes. The remaining constant, represented by “α” in the price equations above can be considered the base unit common to each of the sales transactions in the data set. By including a time “dummy” variable in the price equation for each time period in the data set, the resulting coefficient βt can then represent the amount that prices in timet have shifted since the base period (t=0)1. After applying the price shifts for each period as indicated by the β coefficients, if all other attributes remain equal (which in the case of this study can just be left out), the result is a true housing price index – an estimate of the price of the base unit of measurement over time. 3.2.1 Price Data Collection The Park City price indices in this research project are constructed using the hedonic regression analysis methodology described above. The problem with estimating an effective hedonic equation, however, is that large amounts of data are needed to help control for the many different attributes that effect price. Data on property sales over a 30-year time period is difficult to find, and the data that is found is not likely to include many of the observable attributes that are needed to effectively predict price. Some variables, such as quality and location which are both very influential pricing attributes, are quite subjective, and would be very difficult and time consuming to quantify. There is a separate pricing methodology known as the repeat sales model which is similar in that it tracks sales transactions over time, but only examines transactions in which the same house has been sold at least twice over the time period being researched2. This methodology eliminates the need for detailed quality attributes because it tracks price movements of the exact same asset. 1 DiPasquale, Denise and William C. Wheaton. Urban Economics and Real Estate Markets. Prentice-Hall, Inc. , 1996. 2 DiPasquale, Denise and William C. Wheaton. Urban Economics and Real Estate Markets. Prentice-Hall, Inc. , 1996. 14 To help control for quality and location attributes in a manner similar to the repeat sales methodology, data was collected only for sales transactions of condominium units in large condominium projects located within 0.75 miles of the resort base and at least 25 years old. Unlike the repeat sales methodology, the observed transactions were not necessarily limited to those of units that sold more than once, but because condominium units in a particular complex are so similar, the results are essentially the same. The units in a particular complex all share the same location1 and are constructed at the same time. They are also expected to be of uniform quality and layout when constructed. This eliminates the need to collect data for, and to assign observable values to, unit quality attributes. To select which condominium projects to examine, the Summit County Assessor’s Office provided a list of all condominium projects in the valley, organized by neighborhood. The list indicated the name and address of each project as well as the number of units and the date the project was platted. A number of potential projects were identified near each of the three resorts based on location, age, and number of units. A site visit to each project was then conducted to observe general quality and maintenance of the projects, and to identify which projects could collectively be representative of the market as a whole. 7 projects were selected to represent the Park City Mountain resort, containing a total of 590 units. 10 Projects with 385 units were selected in the Lower Deer Valley market, and 3 projects with 470 units were selected near the base of The Canyons. The state of Utah is classified as a “non-disclosure” state which means that while changes in ownership of a real estate asset is recorded in the deed of registry, and made public, the transaction sales price is not. However, The Summit County tax assessor indicated that their source for transaction data to aid in the appraisal and tax assessment process is the Park City Board of Realtors. The Park City Board of Realtors was founded in 1980 and operates and maintains the Park City Multiple Listing Service (MLS). The Park City MLS is a service that compiles real estate sales data which is made available to members or subscribers to facilitate 1 This method does not account for location differences within a condominium complex nor quality differences that might result over time due to individual unit ownership. 15 sales and track information. The MLS tracks all real estate transactions in the area that have been listed through the Board of Realtors, estimated to be 90% of all housing transactions. The database includes sales transaction data that is catalogued in computerized format back until 1993. Transactions prior to 1993 have been recorded and kept in old MLS listing booklets. The MLS data is not intended for public use, but The Park City Board of Realtors agreed to supply the data for this study due to its academic nature. A digital file, which contained the data for approximately 1700 sales transactions from the chosen condominium projects from 1993 to 2010, was provided. Additionally, the Board of Realtors provided access to the historical MLS Listing Booklets from which an additional 1300 sales transactions were manually recorded. Table 1 summarizes the data collected. Sales Transaction Data Location Park City Deer Valley Canyons Data Observed: Condo Projects Units Observed Transactions 7 10 3 590 385 470 1,143 957 896 Price Square Footage Bedrooms Bathrooms Date of Sale Condominium Complex Table 1 – Sales transaction data 3.2.2 Index Construction To construct the price index the price per square (PSQFT) for each sale in the data set is first calculated to be used as the dependent variable representing price in the hedonic equation. The independent variables used to estimate price are the remainder of the numerical data collected, including square footage, number of bedrooms, and number of bathrooms. These attributes help control for differences between each of the unit types in a particular condominium development. A separate dummy variable is used for each condominium project, which controls for variations in location, quality, and age. The coefficient (β) for these dummy variables will each represent the estimated difference between the price of units in the corresponding condo project and that of the base project. A time dummy is also used for each of the 29 years following the base year of 1981 to indicate the shift in price over time. The regression analysis results indicate that the estimated hedonic price equation is quite effective, with an R square (R2) of 0.915 for the Park City Resort index. R2 is a statistical measure that in this case indicates that 91.5% of the price can be explained by the various 16 independent variables that have been included in the equation. An R2 of 0.91 is quite high for an equation predicting price per square foot as opposed to total sales price.1 The coefficients for each of the numerical variables are all statistically significant and coefficients all have the right signs. For example, the square footage variable in the Park City linear price equation has a negative coefficient (β) of -.053. This can be explained by the concept of diminishing marginal utility, in that an increase in square footage, while expected to increase the overall value of the house, will do it at a decreasing rate. A larger house, while worth more overall than a smaller one, will actually have a smaller price per square foot, all else equal. The variables of Bedrooms and Bathrooms, on the other hand, both have positive coefficients, indicating that an increased number of each of these attributes has a positive effect on price. The dummy variables for each of the apartment complexes are all significant, and prove to make sense in that the luxury condo complexes have a higher coefficient indicating a greater implicit value over the price of the base project. The coefficients for the time variables are very interesting and are reflected in the Price Index in Table 2 and illustrated in Figure 2. Full regression results for each price equation can be found in the Appendix. It should be noted that an hedonic equation was estimated in both linear and log form. The resulting price indices are nearly identical. The remainder of this chapter will be examining the linear price equations. See the appendix for an illustration depicting the Linear and Log equation for prices in Park City. 1 Wheaton, W. C., “Resort Real Estate: Does Supply Prevent Appreciation?” Journal of Real Estate Research ,Vol 27, 2005. 17 Hedonic Price Index Nominal $ / SF Park City Deer Valley Canyons 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Annual Increase (Observations) (R*2) $162.44 $134.58 $144.45 $138.71 $116.03 $108.26 $100.62 $102.72 $111.55 $120.37 $128.95 $123.64 $133.49 $158.30 $192.76 $230.56 $239.40 $246.42 $245.81 $232.88 $220.37 $221.05 $230.90 $261.44 $339.61 $506.63 $516.47 $468.92 $384.26 $341.64 2.60% 1141 0.9153 $195.41 $187.20 $186.33 $169.97 $154.71 $151.90 $129.20 $139.15 $144.22 $151.17 $146.27 $147.31 $156.79 $165.65 $186.89 $226.95 $237.36 $246.90 $245.41 $235.22 $233.52 $224.29 $233.73 $239.98 $306.93 $448.33 $457.05 $353.83 $341.43 $384.41 2.36% 957 0.8969 $122.88 $128.65 $107.80 $102.92 $99.78 $78.09 $80.86 $82.57 $87.42 $92.25 $97.31 $99.85 $106.18 $132.65 $157.13 $179.24 $186.02 $202.84 $195.44 $185.49 $189.91 $185.20 $175.33 $196.67 $276.18 $371.35 $384.13 $313.08 $239.74 $228.45 2.16% 896 0.9378 Table 2 – Hedonic Price Indices 18 Park City Real $ / SF Deer Valley Canyons $162.44 $126.77 $131.83 $121.35 $98.03 $89.79 $80.51 $78.93 $81.77 $83.71 $86.06 $80.11 $83.97 $97.10 $114.97 $133.58 $135.59 $137.42 $134.12 $122.93 $113.11 $111.69 $114.07 $125.81 $158.07 $228.44 $226.42 $197.98 $162.81 $142.34 $195.41 $172.71 $165.75 $145.12 $127.58 $120.58 $101.08 $104.63 $103.61 $103.23 $94.54 $92.80 $95.66 $98.57 $108.18 $127.88 $129.79 $132.92 $129.95 $121.23 $116.03 $110.18 $111.91 $112.73 $140.03 $196.70 $196.71 $145.84 $140.68 $154.34 $122.88 $118.69 $95.90 $87.87 $82.28 $61.99 $63.26 $62.09 $62.80 $63.00 $62.90 $62.90 $64.78 $78.94 $90.95 $101.00 $101.72 $109.20 $103.49 $95.60 $94.36 $90.98 $83.95 $92.39 $126.00 $162.92 $165.32 $129.04 $98.78 $91.72 -0.45% -0.81% -1.00% Nominal vs. Real Price Index $600.00 $500.00 Price $/SF $400.00 $300.00 $200.00 $100.00 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 $0.00 Park City Park City Deer Valley Deer Valley Canyons Canyons Figure 2 – Nominal Price Index as compared to Real Price Index and New Construction 3.3 Index Analysis The results of the hedonic price indices are quite interesting. At first glance at nominal prices the cyclical nature of the real estate market is revealed. The three property price indices essentially follow identical patterns over the thirty year period. Nominal Prices increase 110% over this period, but this is only a 2.5% annual increase. What is most surprising to observe is that nominal prices for all three markets decreased approximately 35% between 1981 and 1987 and didn’t recover to 1981 prices until 1995. It is also noticed that the overall market, similar to that of the rest of the country, experienced unprecedented nominal price growth of over 100% between 2003 and 2007, and has rebounded sharply with 30%-40% decreases since then. To enable closer examination, Figure 3 below illustrates the real price indices together with annual building permits 19 Real Price Index $250 700 600 $200 $150 400 300 $100 Units Real Price 500 200 $50 100 PC Bld Prmt PC Real $/SF DV Real $/SF 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 1984 1983 1982 0 1981 $0 CN Real $/SF Figure 3 – Real Price Indices and New Construction After adjusting for inflation we can take a closer look at real price changes over the last 30 years. Real prices have failed to keep up with inflation since 1981, having decreased by a total of 12% in the Park City resort market, 18% at Deer Valley, and 25% at the Canyons. What is most astonishing is the drastic change in price between 1981 and 1987. Prices fall across the board to approximately 50% of their 1981 values where they basically remain constant until 1992 and 1993. At this point prices begin to recover, but they don’t reach 1981 levels again until the frenzy of the most recent housing bubble in 2006. Deer Valley prices, in fact, at the peak of the market in 2006, only exceed 1981 prices by 3%. While we don’t have data for prices in park city before 1980, it can be derived that 1981 was the tail end of an inflated real estate cycle similar to many markets across the country. The downward response to these inflated prices was likely exacerbated by record construction numbers in 1981 as well as an abnormally high inflation rate of 6%. As prices hit bottom in 1987 construction comes to a standstill, and only gradually picks up the next couple of years. Any price recovery at this point is stymied by four straight years of 4%-5% inflation. In 1992 prices begin to rise again increasing 72% through 1998. It is interesting to note that 1995 is the year that it was announced that Salt Lake City would host the Olympics of 2002. 20 This announcement likely contributes to two straight years of 16% annual price increases in Park City through 1995 and 1996. It likewise contributes to two years of extremely high construction in 1996 and 1997. In 1998 prices level out and soon take a negative turn as the dot.com bubble bursts. Real Estate prices decrease 20% through 2002 before the real estate bubble causes a 104% price increase from 2002 – 2006, followed by a price drop of nearly 40% through May of 2010. 3.3.1 Comparison of Park City to Deer Valley Comparing Deer Valley to Park City has provided some interesting observations1. The price indices reveal that during market downturns the Park City and Deer Valley real estate prices have reacted nearly identically, however during periods of price growth the Park City market has repeatedly outperformed Deer Valley. See Table 3 for details. Park City Deer Valley Cyclical Comparison % Change in Price Index in each cycle 1981-1992 1992-1998 1998-2002 2002-2006 2006-2009 -51% 72% -19% 105% -29% -51% 44% -18% 78% -28% 2010 -13% 11% Table 3 - Cyclical price comparison, Park City vs. Deer Valley From 1981 – 1992 both indices reflect a 51% decrease in price. However, the following growth period from 1992-1998 results in a 72% increase in Park City prices but only a 44% increase for Deer Valley. Figure 4 illustrates this observation. Along the same lines Park City and Deer Valley experience a 19% and 18% price reduction from 1998 – 2002. However, price increases in the real estate boom of 2002-2006 are observed to be 105% for Park City compared to 78% for Deer Valley. Finally, the indices indicate a similar price decrease of 28% and 29% until 2009, before the sudden 11% price increase in Deer Valley during the first 5 months of 2010. 1 The focus of this research paper is the Park City Resort market, with some comparisons to the Deer Valley market. The Canyons is located outside of Park City limits, and therefore was not able to be examined relative to the housing stock. The Canyons Price Index was included in this Chapter as a comparative measure reflecting cycles across separate nearby markets, but will not be examined further. 21 200% 180% 160% 140% 120% 100% PC 80% DV 60% 40% 20% 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 0% Figure 4 – Percentage Change in Price Index – Park City vs Deer Valley 1992-2010 The difference in appreciation between the two resorts might be attributed to a slight positive location bias in the Park City index due to a supply constraint in its immediate surrounding area. The Park City Resort market has been around longer, is located adjacent to historic downtown Park City, and is relatively mature. It is difficult to find development sites comparable to those upon which the Park City price index is based. In fact while there has been considerable development in Park City proper, few new projects have been completed adjacent to the Park City resort in the time period of our study. On the other hand, in Deer Valley, which is considered the more luxurious and expensive location, the resort opened in 1981, and most of the development surrounding the resort has taken place after this date. While the condominium projects examined in the Deer Valley study are all excellently located in Lower Deer Valley, there have been a number of new developments with comparable locational value since 1981. In fact, the neighborhood known as Upper Deer Valley has essentially been developed in its entirety since 1981, and would probably be considered a higher-end location. It could be that new development surrounding Deer Valley has actually prevented price appreciation in the area from keeping up with the neighbor resort. It would be interesting to break down the new construction numbers over the study timeframe into submarkets to examine the more immediate effects of new supply on prices between the resorts. 22 3.4 Conclusion In spite of the 12% overall decrease experienced in Park City property values over the past 30 years the price index actually reveals a positive linear trend. A final observation to consider is the 70% real price increase in Park City from 1990 to 2010. This 20-year period covers 2 full real estate cycles, measured from trough to trough (assuming that prices have neared the bottom of the current downturn, which may not be the case). The 2.69% annual increase in real price throughout this time period is a healthy increase and encourages the likelihood of future price appreciation in the Park City market. 23 4.0 Time Series Analysis To study the determinants of movements in the property price indices a time series analysis is performed. This is done by using multiple regression analysis to estimate hedonic equations which predict new building supply, skier days (demand), and price. These three equations are then used to create an econometric model which is classified as a conditional Vector Autoregression Model (VAR). A VAR examines the evolution and interdependencies between multiple time series of different variables. In this study the interdependent variables in the VAR are Price and Stock, while Skier Visits is observed as an exogenous demand variable. To complete this model a time series was collected for each of the following variables: TIME SERIES DATA Included in Model: Variable Stockt PCSkiDayt SNWFt PRPricet DVPricet Permitt USINCt Definition Stock of housing in Park City Municipal Skier Visits in Park City Area Park City Snowfall Price Index for Park City and Deer Valley Price Index for Deer Valley New construction permits for Park City Municipal United States real disposable income per capita Examined but disregarded from model due to insignificane: Ratet Interest Rate RMINCt Rocky Mountain disposable income per capita UTINCt Utah disposable income per capita UTSKiDayt Skier visits in Utah USEMPLt US Employment Table 4 – Time series data, variables and definitions 4.1 Park City Skier Demand Skier Visits has been identified as a good measure of ski resort housing demand, due to the fact that it represents a number of potential of renters and buyers of housing in resort areas. 24 The National Ski Areas Association (NSAA) defines a skier visit as “one person visiting a ski area for all or any part of a day or night for the purpose of skiing.”1 Annual skier visits is a measure of the number of skier visits in a specified geographical region per ski season, which is generally November – April/May. We were not able to track annual skier visits for the individual resorts that we are examining, due to the fact that resorts keep that information private. Ski Utah is a trade organization that promotes the Utah ski industry and publishes annual skier visits in the state of Utah dating back to 19802. The Park City Chamber of Commerce and Visitors Bureau also publishes annual skier visits for just the park city area, which consists of Park City Mountain Resort, Deer Valley, and The Canyons3. While the Chamber of Commerce had only published the skier data back to 1990, the staff provided the remaining data which dated back to 1983. Skier Days - Park City & Utah 4,500,000 4,000,000 Skier Visits 3,500,000 3,000,000 2,500,000 2,000,000 Park City Area 1,500,000 Utah 1,000,000 500,000 2010 2007 2004 2001 1998 1995 1992 1989 1986 1983 1980 0 Figure 5 – Skier Visits (see table of data in Appendix) The data reflects that skier days in both the Park City area and overall Utah market have followed similar cyclical patterns, with considerable growth over time. Park City skier days increased at a greater rate with a cumulative increase of 142% since 1983 compared to 75% for Utah skier days. The three Park City resorts accounted for 43% of Utah skier visits in the 2009/2010, compared to 31% in 1983. Growth in skier days in both Park City and Utah has 1 (NSAA) National Ski Areas Association. "Estimated U.S. Ski Industry Visits by Region 1978/79 - 2008/09." 2009. www.nsaa.org. 1 6 2010 <http://www.nsaa.org/nsaa/press/historical-visits.pdf>. 2 Ski Utah. "Utah Skier Days Table." 24 6 2010. www.skiutah.org. 24 6 2010 <http://www.skiutah.com/media/story_starters/utah-skier-days-table>. 3 Park City Chamber and Visitor's Bureau. "Economic and Relocation Package - Park City History." 2010. ParkCityInfo.com. 5 June 2010 <http://www.parkcityinfo.com/docs/PARK_CITY%20HISTORY%202009.pdf>. 25 considerably outpaced that of the Rocky Mountain and national ski industries. See Figure 6 below to compare growth. Skier Days % Growth 180.00% 160.00% 140.00% 120.00% 100.00% US 80.00% Utah 60.00% Park City 40.00% Rocky Mtn 20.00% 0.00% 2009 2007 2005 2003 2001 1999 1997 1995 1993 1991 1989 1987 1985 1983 -20.00% Figure 6 – Skier Day Growth, National Comparison 4.1.1 National Economic Data Series To study the determinants of movements in the Skier Visit series, employment and income data were collected at the state, regional, and national level. The economic variable that proves most influential to the Utah ski business is real disposable income per capita1. Subsequently, data series of disposable income per capita of the Utah, Rocky Mountain, and the United States regions were all examined closely as part of various estimated equations predicting skier days. Not surprisingly, while all three series are observed to be influential the most effective economic determinant of the Utah and Park City ski business proves to be nationwide U.S. Disposable Income Per Capita (USINC). Equations were estimated using multiple variations of contemporaneous, first, and second order lags. The most effective Skier Day equation was estimated using the first order lag of USINCt-1. This is understandable, as Park City is a destination resort that depends largely on customers that visit on an extended vacation, from all over the country. Such vacations are generally planned far enough in advance that disposable income levels from the previous year appear to be the greatest determinant for the number of 1 Unemployment rates and overall employment and income levels were also examined. Data Source: Bureau of Economic Analysis, US Department of Commerce, March, 2010. 26 visits. The fact that a second lag doesn’t significantly improve the equation implies that disposable income growth generally stimulates a permanent increase in skier days. Figure 7 below reflects Skier days compared to disposable income levels since 1983. While it is difficult to see the effect that minute changes in disposable income have on the fluctuations of skier visits, the graph does reflect the long-term growth pattern of both series1. The next figure (8) reflects percentage growth of skier days compared with percentage growth of USINCt-1. This figure illustrates that a small increase in the growth rate of U.S. disposable income has a significant effect on the growth of Park City skier visits. The estimated equation predicting skier days is labeled Equation 2 in the next section. 200 1000.0 180 900.0 160 800.0 140 700.0 120 600.0 100 500.0 80 400.0 60 300.0 40 200.0 20 100.0 SnFall PC Area Ski Days 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 1984 0.0 1983 0 U.S. Income Figure 7 – Park City Skier Days, U.S. Disposable Income, Snowfall 1 The average annual growth rate of U.S. real disposable income per capita is 1.36% 27 Snowfall(inches) US Income Skier Days (000's) Park City Area Skier Days SnFall Ski Visit Growth 2010 2009 2008 2007 2006 2005 2004 2003 2002 -20% 2001 100.0 2000 -15% 1999 200.0 1998 -10% 1997 300.0 1996 -5% 1995 400.0 1994 0% 1993 500.0 1992 5% 1991 600.0 1990 10% 1989 700.0 1988 15% 1987 800.0 1986 20% 1985 900.0 1984 25% 1983 Snowfall Inches Skier Day % Growth 1000.0 U.S. Income Growth Figure 8 - Skier Day Growth, U.S. Disposable Income growth, Snowfall 4.1.2 Annual Snowfall Another determinant of skier business is the amount of snow that falls in a particular area over a season. While most Utah resorts have implemented artificial snow making systems – including all three Park City resorts - annual snowfall is still reported by all resorts as part of their marketing packages, as it is widely thought that the amount of snowfall affects the overall ski experience. The IBIS World Ski Industry Report indicates that ski resorts focus on two separate customer bases: the local skier market, and the destination skier market. The local market is largely influenced by both ski conditions and travel time, while the destination skier market is influenced more by the entire vacation experience (nightlife, lodging, restaurants, etc) 1. Local business therefore varies greatly due to unpredictable snowfall and other weather conditions. Resorts try to neutralize the volatility caused by weather conditions by marketing season passes, which are sold before the season begins, to the local communities. Resorts also combat the unpredictability of the weather by making artificial snow. All three Park City resorts, particularly 1 IBIS World. "Ski Resorts in the US, IBIS World Industry Report 71392." January 2010. IBIS World. 6 June 2010 <http://www.ibisworld.com/industryus/default.aspx?indid=1653> 28 Deer Valley, keep most of their beginner and intermediate runs covered and well-groomed to ensure a positive skiing experience, despite the lack of any recent snowfall. A series of data indicating annual snowfall at Park City Mountain Resort dating back to 1980 was provided by the media office of the resort. This data is considered to be representative of the rest of the ski market due to proximity of the other resorts and common weather patterns. The series indicates total snowfall over each season as measured at the summit of Jupiter bowl, which is the point of highest altitude at Park City resort and the area that receives the most amount of snow. The snowfall data, depicted above in Figures 6 and 7 along with skier days and income levels, is quite volatile with a low annual snowfall of 169 inches in 1981, a high of 512 in 1993, and an average annual snowfall of 365 inches. The ski visit equation was estimated using various lags of this series as well to determine if the snowfall of previous years might have an effect on the current year ski business, but the contemporaneous variable was the only one with any significance. It is interesting to observe in Figure 7 the effect that snowfall has on the growth in skier days. Almost without exception the years with the largest amounts of skier day growth are years reflecting both an increase in income growth and above average snowfall. 4.1.3 Skier Visit Equation The results of the regression analyses to predict determinants of skier visits in the Park City ski area are depicted below as Equation 2. A full regression summary can be found in the appendix. PCSkiDayt = -1099974 + 0.4141PCSkiDayt-1 + 563.26SNWFt + 124.59USINCt-1 (t Stat) (-4.1) (3.05) (3.16) (4.07) R2 = .955, N = 27 (1983-2010) Equation 2 – Park City Skier Days While snowfall definitely does have a determining effect on the amount of skier visits in the region, disposable income proves to have greater long term effects, as indicated, in part, by the greater t-stat of 4.07. 29 The equation reflects that a one year positive increase of annual snowfall to 500 inches (nearing the 30 year record of 512 inches), would result in a 4.35% increase in skier days for that year, which is a considerable effect. However, the following year, as snowfall drops back to average levels, the amount of skier visits drops back to just 1.8% greater than the level prior to the shock, and within a few more years any positive effect on skier days has essentially disappeared. On the other hand, the effect that change to disposable income has on skier days is a bit different, in part because disposable income experiences growth fairly continuously. In the 30-year time examined in this study, the average growth rate of real disposable income has been 1.36%. The series only reflects negative annual growth 5 total years throughout that time. An increase in the growth rate of disposable income from 0% to 1% for one year results in a 1.1% increase in skier days that first year. If the growth rate is reset to 0 after the first year, the impact of that one year of growth is still reflected in the number of skier days which increases through year 8 before it holds steady at a 1.88% increase. If the 1% increase in the growth remains permanent, the number of skier days continues to grow annually, reaching an increase of 18.2% in year 10. The results of the estimated skier visit equation verify that the ski business of a destination resort area, such as Park City, Utah, is most heavily influenced by the national economic factors, such as U.S. disposable income per capita1. As visitors from around the country are a large part of the Park City business, and generally plan a trip long before the snow season has begun, snowfall has less of a long-term effect on business. 4.2 Supply As described in Chapter 3, the supply variable used to examine the fluctuations in the price series is stock, which in this study is defined as the number of dwelling units within the municipal boundaries of Park City. The stock series can be defined by the following equation: 1 It is interesting to note that the skier day equation predicting Utah Ski Visits had similar results, except that snowfall has a larger significance relative to income growth. This reflects the fact that compared to the rest of the Utah resorts Park City is more of a national destination. The Park City skier day equation is used in this analysis as it is a more significant determinant of Price. 30 Stockt = Stockt-1 + Permitt-1 Equation 3 – Stock1 The equations predicting construction permits were estimated using different lags of the Price(Park City and Deer Valley), Stock, and Permits data series. Interest rates and skier visits were also included in the exercise as exogenous variables, but neither proved to provide any significance to the equations. The resulting permit equations are as follows: Permitt = 197 - 0.0664Permitt-1 – 0.0487Stockt-1 + 2.274PCPricet (t Stat) (2.18) (-0.378) (-2.936) (3.117) R2 = .334, N = 29 (1981-2010) Equation 4 – Permit Equation (Park City Prices) Permitt = 45.08 – 0.044Permitt-1 – 0.023Stockt-1 + 2.316DVPrice (t Stat) (0.363) (-0.243) (-1.632) (2.759) R2 = .291, N = 29 (1981-2010) Equation 5 - Permit Equation (Deer Valley Prices) While the permit equations establish that construction permits can be hard to predict, both equations illustrate that price clearly has the largest effect on new construction. A 5% increase in price, for example, would cause a 12% increase in construction permits. However, this 12% increase in permits represents an overall stock increase of only 0.2%. Figure 9 below illustrates the construction permit series data and its relationship to the price index. While the amount of annual permits fluctuates considerably the general correlation with price fluctuations is reflected quite clearly. The negative coefficients of the lagged Permit and Stock indexes counteract increases influenced by price over the next two years, but the effect is minimal. The effects of the variables in the 1 The stock equation assumes that additions to stock are permanent, essentially ignoring demolition, which is assumed to be inconsequential in the Park City market. 31 permit equation will be examined further as part of the complete forecasting model discussed in the next chapter. It should be noted that the permit equation examining the effects of Park City prices is noticeably more effective than that of Deer Valley Prices with a higher R2 and more significant variables. Subsequently the Park City equation is examined more fully in the forecasting model detailed in this study. Price vs. Const. & Ski Days 250.00 900 800 Price($) & Skier Days (10,000s) 200.00 700 600 150.00 Units 500 400 100.00 300 200 50.00 100 0.00 0 Stock (10%) Permits PC Price Ski Days Figure 9 – Park City Prices vs. Construction & Skier Visits 4.3 Price To examine the determinants of fluctuations in the price series, various equations were estimated using different lagged values of price, skier days, stock, and interest rates as the independent variables. It was surprising to observe that interest rates produced little significant effect on fluctuations of the Price series. This suggests that many second homes in the Park City market 32 are purchased with cash, an argument supported in part by the National Association of Realtors 2009 Buyers Survey, which indicates that 3 in 10 vacation homes were purchased with cash1. The results of the various equation estimates also indicate once again that second order lags provide little significance in the prediction of price fluctuation. In fact the most effective equation also proves to be the most simple: PCPricet = 5.653 + 0.4972PCPricet-1 + 0.000116PCSkiDayt – 0.0145Stockt-1 (t Stat) (0.517) (3.815) (3.43) (-2.429) R2 = .864, N = 28 (1983-2010) Equation 6 – Park City Price Equation (Time Series) DVPricet = 27.272 + 0.510DVPricet-1 + 0.0000756PCSkiDayt – 0.0103Stockt-2 (t Stat) (1.826) (3.787) (2.455) (-1.762) R2 = .743, N = 28 (1983-2010) Equation 7 – Deer Valley Price Equation (Time Series) The above equations depict that the price of Park City real estate is determined by SkierDays and Stock, supporting, quite simply, one of the basic principles of economics: that price is a function of supply and demand. While the price equation can be used to derive single-year calculations of supply elasticity, the long run effects of changes to these variables cannot be determined by this equation alone. This is due to the interdepency between the price and stock variable. But the price and stock equations together, combined with conditioning demand equation (skier days) will comprise the forecasting model which will enable the examination of the long run effects of variable fluctuations. Figure 8 above also illustrates the Park City price series in relation to the number of skier days, stock, and construction permits (a measure of the change in supply). It can be observed that 1 National Association of Realtors. Second Homes: Talking Points. 10 March 2010. 6 July 2010 <http://www.realtor.org/press_room_secured/public_affairs/tpsecondhomes>. 33 price follows the general growth trend of skier days, but that the growth in price is occasionally reversed, often in response to increased construction. Comparing the Deer Valley equation to the Park City equation reveals that both markets behave similarly. The Park City model, however, appears to be slightly more effective, with a higher R2 and more significant variables. The forecasting model examined in the following chapter will therefore be constructed with the Park City price and stock equations. This price equation combined with the other equations predicting permits, stock, and skier visits (demand) make up the Vector Auto Regression forecasting model that is used to forecast levels of each variable, as well as to examine behavioral patterns caused by various shocks to the system, as discussed in the following chapter. 34 5.0 Forecasting Model The equations derived in the time series analysis as detailed in chapter 4 are used to construct an econometric model that predicts the behavioral relationship between price, stock (as determined by new construction), and demand in the park city market based on the behavior of those variables over the last 30 years. In this particular model price and stock are the endogenous variables that are interdependent while the annual skier days is used as the conditioning variable that represents demand. The purpose of this model is to predict the reactions of the endogenous variables - price and stock – to fluctuations in either of these variables or the conditioning variable of skier days. As described in Chapter 4 fluctuations in skier days can be determined by snowfall and growth of real U.S. disposable income per capita, two completely exogenous variables. Therefore demand shifts in this model can be implemented simply by changing either of these two exogenous variables. The model is illustrated below in Figure 10. PCSkiDayt = -1099974 + 0.4141PCSkiDayt-1 + 563.26SNWFt + 124.59USINCt-1 (t Stat) (-4.1) (3.05) (3.16) (4.07) R2 = .955, N = 27 (1983-2010) Equation 2 – Park City Skier Days Stockt = Stockt-1 + Permitt-1 Equation 3 – Stock Permitt = 197 - 0.0664Permitt-1 – 0.0487Stockt-1 + 2.274PCPricet (t Stat) (2.18) (-0.378) (-2.936) (3.117) R2 = .334, N = 29 (1981-2010) Equation 4 – Permit Equation (Park City Prices) PCPricet = 5.653 + 0.4972PCPricet-1 + 0.000116PCSkiDayt – 0.0145Stockt-1 (t Stat) (0.517) (3.815) (3.43) (-2.429) R2 = .864, N = 28 (1983-2010) Equation 6 – Park City Price Equation (Time Series) Figure 10 – Econometric Forecasting Model 35 To illustrate the use of this model a forecast has been created based on a realistic demand scenario, with average annual snowfall and average growth of disposable income. Positive and negative demand “shocks” are then applied to the model, to create optimistic and pessimistic forecasts. The reactions of the variables to these shocks relative to the base case are then analyzed and the long run price elasticity of supply is calculated. 5.1 Base Forecast To create the realistic forecast of price, construction, and skier days in the Park City market, the average annual snowfall of 365 inches and the average income growth rate of 1.36% were applied to the skier day equation for each year. The resulting 15-year forecast, compared with the actual values since 1980, can be observed in Figure 10 below. Table 4 below summarizes the forecast. Price vs. Const. & Ski Days 700 600 250.00 500 200.00 Units 400 150.00 300 100.00 200 50.00 100 0.00 0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 Price($) & Skier Days (10,000s) 300.00 Permits Pmt Forecast PC Price Ski Days Price Forcast SD Forecast Figure 11 – 15-Year Forecast – Realistic Demand Scenario 36 Year Average 1981-2010 Real Price Total Stock 125.39 5,281 Annual Permits Skier Visits (Thousands) 215.5 1,180 Forecast 2010-2025 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 Total Growth Ave. Growth Snowfall (in) 365 365 $142.34 $156.54 $164.66 $170.85 $176.59 $182.32 $188.16 $194.08 $200.07 $206.10 $212.17 $218.25 $224.37 $230.51 $236.68 $242.90 70.65% 3.63% 8,362 8,471 8,610 8,760 8,916 9,078 9,245 9,417 9,594 9,776 9,963 10,154 10,350 10,550 10,754 10,962 31.09% 1.82% 109 139 150 157 162 167 172 177 182 187 191 196 200 204 208 212 95.12% 4.56% 1,734 1,726 1,749 1,784 1,826 1,870 1,916 1,962 2,010 2,058 2,107 2,157 2,207 2,258 2,309 2,362 36.20% 2.08% Dis. Inc. Growth Disposable Income 1.36% 12,257 1.36% 377 365 365 365 365 365 365 365 365 365 365 365 365 365 365 365 1.36% 1.36% 1.36% 1.36% 1.36% 1.36% 1.36% 1.36% 1.36% 1.36% 1.36% 1.36% 1.36% 1.36% 1.36% 15,064 15,269 15,478 15,689 15,903 16,120 16,340 16,563 16,789 17,018 17,250 17,485 17,724 17,966 18,211 18,459 22.54% 1.36% Table 5 – Forecast 2010-2025, Realistic Scenario The forecast predicts that prices start to recover quickly with a 10% increase in 2011, and a 5.19% increase the following year. Prices continue to increase, at slightly decreasing rates, through the 15 year period for a total increase of 70.65% and an average growth rate of 3.6.%. This forecast is consistent with the trailing 20-year trend, and is likely stimulated in the short run by the 2010 increase in skier days. New construction permits are predicted to drop considerably from the unusually high number of 289 permits issued in 2009 to 109 permits in 2010. From there they increase a considerable 29% to 139 permits in 2011, 8% in 2012, and continue to respond to rising prices with annual increases, at decreasing rates through 2025. Skier days in park city are predicted to slightly decrease (-0.5%) in 2011, after which they begin a steady increase between 2% and 2.5% each year for the remainder of the forecast. These forecasted numbers appear reasonable with the implemented exogenous variables of average snowfall and income growth, suggesting a quick recovery in the short run and steady 37 growth in the long run. This forecast also gives us a base case against which we can compare reactions to positive and negative demand shocks. 5.2 The Reaction of Forecast to Temporary Shocks One of the benefits of the forecasting model is that once a base case is established the reactions to demand shocks can be examined more closely relative to that of the base case scenario. To do this an impulse response function is created that measures the percentage change in forecast from the base case as a result of the positive or negative demand shock. The traditional impulse response function measures reactions within a system caused by a transitory, or in this case a one-year shock to the system. In this section the system’s response to multiple transitory shocks will be observed. 16% 14% 12% 10% 8% Price 6% Skier Days 4% Stock 2% Permit 0% -2% -4% 0 5 10 15 20 25 Figure 12 – Impulse response relative to base forecast, one-year demand shock of 500” of snowfall The above figure illustrates the response of the system to a one-year increase in snowfall from 365 inches to 500 inches. It is interesting to note that the 4% increase in demand caused by the temporary increase in snowfall quickly disappears and the amount of skier days returns to preshock levels. Prices react in a similar manner, increasing 6% but decreasing again as the demand returns to original levels. However, construction reacts to the increase in price adding additional units to the market. Due to the fact that any increases in stock are permanent increases in this model, the increased supply, with no long term change to demand, causes prices to drop slightly below pre-shock levels in year 6. Construction quickly reacts by decreasing, which leads to a slow price recovery, starting in year 9. It should be noted that these percent changes are relative to the base forecast scenario specified in section 5.1. In this case, prices never actually fall below 2010 levels, they just fall slightly below the levels of the base case scenario. For example 38 in 2019, post-shock prices reach $204.96 as opposed to $206.10 in the base case. The behavior exhibited in Figure 13 reflects typical market reactions to temporary increases in demand. Figure 14 below illustrates that a negative temporary demand shock, caused by a decrease in snowfall, reacts similarly to that of the positive shock detailed above. This impulse response illustrates the effects of one year of reduced snowfall in the amount of 200”. 5% 0% -5% Price Skier Days -10% Stock Permit -15% -20% 0 5 10 15 20 25 Figure 13 – Impulse response relative to base forecast, one-year 200” snowfall The data illustrated in Figure 14 suggests that a temporary negative demand shift, caused in this case by reduced snowfall for one year, results similarly to the positive temporary shift in that price is reduced enough to slow construction and as a result, when demand returns to pre-shock levels, price actually appreciates above pre-shock levels. These mirrored reactions to positive and negative temporary demand shifts likely cancel each other out over time. Figure 15 illustrates the response of the forecasting model to a one-year increase in the growth rate of U.S. disposable income, from 1.36% to 2.5%. 39 12% 10% 8% Price 6% Skier Days 4% Stock 2% Permit 0% -2% 0 5 10 15 20 25 Figure 14 – Impulse Response relative to base forecast, one year shock of 2.5% income growth Note that a one-year increase in growth of disposable income causes skier days to increase the following year, but in contrast to the increase caused by snowfall, this shift in demand appears to be permanent. It is interesting to observe that one year of positive income growth causes a 2% growth in skier days, leading to a 4.7% growth in price by year 5. Construction responds similarly with a 10% increase, causing prices to decline slightly, reflecting a 10-year price increase of 4.26% and stock increase of 1.3%. Figure 16 below illustrates the effect of a one-year reduction in income growth from 1.36% to 0.5% 1% 0% -1% -2% -3% Price -4% Skier Days -5% Stock -6% Permit -7% -8% -9% 0 5 10 15 20 25 Figure 15 – Impulse response relative to base forecast, one-year shock of 0.5% income growth It is observed in Figure 16 that the market reaction to a decreased growth rate of income mirrors that of an increased rate. A one year reduction in income growth to 0.5% results in a permanent 40 demand shift, leading to a 5-year price reduction of 3.5%, reduced construction and a 10-year price reduction of 3.24%. 5.3 The Reaction of Forecast to Permanent Shocks Another benefit of the impulse response function is that when used to reflect reactions of permanent demand shocks it can also be used to calculate the long-run price elasticity of supply. Consider figure 17 below. 40% 35% 30% 25% Price 20% Skier Days 15% Stock 10% Permit 5% 0% -5% 0 5 10 15 20 25 Figure 17 – Impulse response relative to base case, permanent shock of 500” snowfall Figure 17 illustrates that a permanent shift in snowfall is predicted to cause a permanent 7% shift in skier days, relative to the base case. This shift causes prices to rise 15% over 5 years, which in turn leads to a construction boom. The increase in stock causes the price increase to temper and the market eventually settles into equilibrium. While this permanent shift is an unlikely scenario, it can be used to represent a permanent shift in demand and the impulse response can be examined to calculate the implied long-run supply elasticity. This impulse response mirrors the typical reaction of any healthy market to a positive shift in demand, reflecting a relatively inelastic supply market. Elasticity of supply is measured as the ratio of % change in supply (stock) to % change in price (price index). It is a measurement that reflects the responsiveness of supply to a change in price. As the impulse response reflected in Figure 13 represents the % increases in price and stock in response to a permanent shift in demand, it is possible to calculate the implied long run supply elasticity. In year 2 after the demand increase, for example, the price increase relative to the base case is 10.28% while the stock increase is 0.23% reflecting an elasticity of .02. By year 10, 41 however, price has increased 13% and stock 4.29% reflecting a long run supply elasticity of .331. An elasticity between 0 and 1 is considered to be relatively inelastic, and prices in a market with a relatively inelastic supply are expected to appreciate any time there is a permanent demand shift. In the Wheaton study which examined Loon Mountain ski resort in New England, this same exercise revealed that Loon Mountain has an elastic supply market that responds so quickly to any price increases that long run appreciation is not to be expected. Figure 18 illustrates the forecast response to a permanent decrease in snowfall to 200”. 5% 0% -5% -10% -15% Price -20% Skier Days -25% Stock -30% Permit -35% -40% -45% 0 5 10 15 20 25 Figure 18 – Impulse response relative to base case, permanent 200” Snowfall The impacts of the permanent reduction in snowfall are similar to those of increased snowfall. Prices depreciate nearly 20% in 5 years, but the reduced construction leads to gradual price recovery. The implied 10-year elasticity of demand is again 0.33, congruent with the positive increase scenario, which is expected, and suggests that the model is functioning properly. A permanent increase in the growth of disposable income reflects a much different response: 1 Long-run elasticity generally increases over short-run elasticity, especially in the housing market, as supply markets take time to react to shifts in demand. 42 200% 180% 160% 140% 120% Price 100% Skier Days 80% 60% Stock 40% Permit 20% 0% -20% 0 5 10 15 20 25 Figure 19 –Impulse response relative to base forecast, permanent 2.5% Income Growth Figure 15 illustrates the system’s response to a permanent increase of income growth to 2.5%. Not unexpectedly the system responds with considerable growth and price appreciation. Relative to the base case, skier days increase 39% over 15 years, price appreciates 70%, and stock grows 15.96%. When compared to Figure 17 above this chart again reflects that the Park City market is much more responsive to shifts in income growth than it is to snowfall. The 10-year elasticity reflected in this scenario is .15, however this number is likely not representative of actual long run elasticity determined with permanent demand shifts, as the demand in this case is increased every year due to the compounding nature of the annual growth rate. The number is likely negatively influenced by the relatively smaller short run elasticity that is prevalent in real estate markets due to the length of time required to get new supply out to market. The response to a permanent downward shift in income growth is illustrated in Figure 20 below. 43 20% 0% -20% Price -40% Skier Days -60% Stock -80% Permit -100% -120% 0 5 10 15 20 25 Figure 20 – Impulse response relative to base forecast, Permanent 0.5% Income Growth As illustrated a permanent decrease in income growth causes a steady decrease in skier days, price, construction, and stock, relative to the base case, which mirrors the reactions of the positive shift of income growth. The 10-year change in price is -31%, and in stock -4.8%, implying a supply elasticity of .15, which again mirrors that of the positive shift. 5.4 Alternative Long-Range Forecasts To provide perspective the econometric model was used to construct long-range forecasts based on best-case and worst-case demand scenarios. Figure 21 below reflects the 15-year forecast of price, construction, and skier days based on the optimistic demand shift caused by a permanent increase in annual snowfall to 500 inches combined with a permanent increase in growth of disposable income to 2.65%. 44 400.00 Price vs. Const. & Ski Days 600 300.00 500 250.00 400 Units Price($) & Skier Days (10,000s) 350.00 700 200.00 300 150.00 200 100.00 100 50.00 0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 0.00 Permits Pmt Forecast PC Price Ski Days Price Forcast SD Forecast Figure 21- Optimistic Forecast – Increased Snowfall In the above forecast, the permanent increases of both snowfall and disposable income cause skier visits to increase from 1.7 million in 2010 to 3.1 million in 2025, reflecting a 82% overall increase and average annual increase of 4.095%. Prices respond early with an 18% jump in 2011, an additional 13% in 2012, and a steady, but slowed increase thereafter for an average annual increase of 6.65%. Construction follows suit with steep increases in the early years, and solid growth thereafter reaching 418 permits in 2025. While the future sustained growth reflected in Figure 21 isn’t likely, it is interesting to observe that growth has occurred at similar rates for various different stretches in the past, and can be quite possible. The forecast representing an assumed worst-case scenario is reflected in Figure 18. 45 Price vs. Const. & Ski Days 700 300.00 600 250.00 500 200.00 400 150.00 300 100.00 200 50.00 100 Units Price($) & Skier Days (10,000s) 350.00 0 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024 0.00 Permits Pmt Forecast PC Price Ski Days Price Forcast SD Forecast Figure 22 – Pessimistic forecast, .5% income growth, annual snowfall 200” Figure 22 above reflects the 15 year forecast of the econometric model based on the demand inputs of 200” of annual snowfall and an annual income growth rate of 0.5%. These negative demand inputs have been selected to illustrate the system’s long run reaction to an assumed worst-case demand scenario1. The system responds first with a 6.75% decrease in skier days in year one, followed by an additional 2.4% decrease in year 2, and a 0.4% decrease in year 3. Thereafter the ski business increases at very low rates. Prices follow demand with 4 straight years of depreciation, but a dramatic decrease in construction permits, which reaches a low of 58 in 2016, causes prices to start to recover slightly in the same year. Surprisingly, over the 15-year pessimistic forecast prices actually end up with a 0.2% overall growth, indicating that, in almost all cases, prices are likely to increase in the mid-term. 1 Interestingly, it has been estimated that global warming, if not controlled, could result in average snowfall decreasing to amounts along these lines. Additionally, the Republican Party has expressed that excessive national debt could drag down long term national growth to 0.5% annually.(Wheaton) 46 5.4 Forecast Conclusion The results of the forecast exercises indicate on all accounts that the Park City second home market is a well functioning market. While transitory positive demand shifts can result in overbuilding and reduced prices, long-run reactions of price and construction to permanent shifts in demand repeatedly reveal a relatively inelastic supply market. As such, any permanent positive shifts in demand should result in price appreciation. The forecasts also suggest that, considering recent market activity, prices should appreciate in coming years in all but the most pessimistic scenarios. 47 6.0 Conclusion The purpose of this research project is to examine the market pricing behaviors of second homes in the ski resort market of Park City, Utah. To accomplish this, in order to track true price appreciation over time a real price index was constructed from 1981 to 2010 for 3 separate localized markets. The resulting price indices reveal a history of cyclical price movements, and surprising long-term price depreciation of 12% to 25% between 1981 and 2010. To determine the causes of the cyclical movements in the price indices, time series regression analysis was performed, and a model was created to predict market behaviors based on past levels of price, construction, and skier days. The results of this exercise reveal that the number of annual skier days in the area is an effective representative of demand, and that the local ski business has a considerable effect on real estate prices. Additionally, it is revealed that the area’s ski business is largely affected by the health of the national economy, reflected specifically by U.S. disposable income per capita. The national economy appears to have more of an effect than the local and regional economy, which is congruent with the resort town’s claim to be a national ski destination. This conclusion is also supported by the fact that the national economy has a greater effect on ski business than annual snowfall. The analysis concludes that despite the thirty year decline in real prices, the Park City resort market behaves as a well functioning, healthy market. The model indicates that while increases in prices do stimulate new construction, the growth in the total number of dwelling units reveals a relatively inelastic supply market. This suggests that any growth in demand should be accompanied with long-term price appreciation. To illustrate the utility of the model, market forecasts based on varied levels of future snowfall and U.S. disposable income levels are performed. The resulting forecasts indicate that except in the most pessimistic cases, prices in Park City should experience healthy appreciation in the near to mid future. 48 As an aside interest the data indicates that prices at Deer Valley, widely considered the more luxurious and expensive of the resorts, did not perform any better than those at Park City. In fact, data suggests that price appreciation at Deer Valley might be curtailed due to a slightly more elastic supply than Park City, which is more fully developed. Both markets, however, behave similarly and can expect to experience price appreciation unless the market changes drastically. The question remains, however: if prices can be expected to appreciate in Park City, then why, over 30 years, have they decreased by 12%? The answer is likely to be, very simply, timing. Real estate is traditionally a cyclical market, and while covering 30 years should help negate any cyclical variations, the time period of this research project happened to begin at the precipice of a very steep and long lived price decline, and to end at the base(to be determined) of an even steeper price decline. Taking a closer look at the conditions in 1981 when this study begins, it is revealed that 1981 experienced the largest decrease in U.S. disposable income per capita over the study period, and the only second consecutive decrease. 1981 also reveals the largest amount of construction permits issued in Park City over that time period (645). Additionally, Park City Mountain Resort reports record low snowfall in 1981 (169”) and although skier visit data is not available for the Park City area in that year, total Utah skier visits decreased 16% that year, the largest decrease in the study period. In summary, 1981, the first year of our price index, experienced extraordinary circumstances that help to explain the subsequent fall in prices, and long recovery period. The information provided through this study, together with that presented in the previous studies which examine the New England and Tahoe resorts, provides insight into the pricing behaviors of different resort communities. The results of these studies help to identify which resort characteristics lead to positive long-term appreciation. Potential second home buyers could use this information to help consider what characteristics to look for in a resort market, and in a location within that market, before making their home purchase. Additionally, Buyers and developers can both use the information in this study to help anticipate pricing shifts, so as to properly time their purchase and/or sale to maximize profits. On another note, city and resort planners could use this information to help develop planning strategies and building regulations to prevent overbuilding and to encourage price appreciation within their markets. 49 As the second home market continues to grow, information regarding the behaviors of the various markets can become more and more useful. It is hoped that the results of this study can help to provide transparency and perhaps lead to further studies of different types of markets in the vacation home industry. 50 Appendices Appendix 1 – Data Park City Housing Supply Year Permits Stock 1980 92 1,897 1981 645 1,989 1982 248 2,634 1983 297 2,882 1984 446 3,179 1985 138 3,625 1986 26 3,763 1987 42 3,789 1988 92 3,831 1989 164 3,923 1990 177 4,087 1991 176 4,264 1992 142 4,440 1993 147 4,582 1994 246 4,729 1995 434 4,975 1996 369 5,409 1997 164 5,778 1998 222 5,942 1999 497 6,164 2000 195 6,661 2001 135 6,856 2002 59 6,991 2003 92 7,050 2004 183 7,142 2005 224 7,325 2006 243 7,549 2007 244 7,792 2008 37 8,036 2009 289 8,073 2010 8,362 Table 6 – Housing Permit and Supply Data 51 Annual Skier Days Year Park City Area Utah 1980 2,055,000 1981 1,726,000 1982 2,038,544 1983 716,468 2,317,255 1984 771,222 2,369,901 1985 789,415 2,436,544 1986 798,311 2,491,191 1987 723,537 2,440,668 1988 767,786 2,368,985 1989 887,314 2,572,154 1990 861,242 2,500,134 1991 943,040 2,751,551 1992 788,830 2,560,805 1993 970,000 2,839,650 1994 992,000 2,808,148 1995 1,137,589 3,113,072 1996 1,055,857 2,954,690 1997 1,211,189 3,042,767 1998 1,204,399 3,101,735 1999 1,203,905 3,095,347 2000 1,158,911 2,959,778 2001 1,278,796 3,278,291 2002 1,161,734 2,984,574 2003 1,343,941 3,141,212 2004 1,418,345 3,429,141 2005 1,608,332 3,895,578 2006 1,715,536 4,062,188 2007 1,746,333 4,082,094 2008 1,871,540 4,249,190 2009 1,645,233 3,972,984 2010 1,734,025 4,048,153 Source: PC Chamber of Commerce Ski Utah Table 7 – Annual Skier Days 52 Income Per Capita Disposable Income Per Capita Year U.S. Utah U.S. RckyMtn Utah 1974 5,707 4,745 5,002 4,528 4,244 1975 6,172 5,180 5,489 5,060 4,693 1976 6,754 5,760 5,965 5,537 5,157 1977 7,405 6,348 6,509 6,017 5,671 1978 8,245 7,054 7,215 6,778 6,291 1979 9,146 7,792 7,952 7,569 6,923 1980 10,114 8,501 8,802 8,443 7,584 1981 11,246 9,374 9,746 9,553 8,325 1982 11,935 9,973 10,410 10,171 8,852 1983 12,618 10,535 11,114 10,706 9,469 1984 13,891 11,431 12,294 11,737 10,325 1985 14,758 12,048 13,008 12,430 10,849 1986 15,442 12,426 13,626 12,609 11,176 1987 16,240 12,729 14,226 12,814 11,392 1988 17,331 13,192 15,271 13,591 11,803 1989 18,520 14,005 16,231 14,378 12,546 1990 19,477 14,913 17,108 15,253 16,149 1991 19,892 15,492 17,578 15,752 16,816 1992 20,854 16,115 18,478 16,609 17,430 1993 21,346 16,756 18,862 17,091 17,925 1994 22,172 17,566 19,550 17,702 18,364 1995 23,076 18,478 20,286 18,351 18,848 1996 24,175 19,529 21,089 19,136 19,159 1997 25,334 20,600 21,941 20,174 20,413 1998 26,883 21,708 23,163 21,698 18,937 1999 27,939 22,393 23,974 22,713 19,488 2000 30,318 24,517 25,955 25,069 21,454 2001 31,145 25,534 26,817 26,474 22,502 2002 31,462 25,648 27,816 27,152 23,061 2003 32,271 25,835 28,829 27,755 23,384 2004 33,881 26,837 30,309 29,133 24,325 2005 35,424 28,617 31,342 30,350 25,555 2006 37,698 30,337 33,174 32,055 26,850 2007 39,392 31,800 34,453 33,180 28,020 2008 40,166 32,050 35,464 33,939 28,585 2009 39,138 30,875 35,553 33,513 28,188 Source: Bureau of Economic Analysis, US Department of Commerce, March, 2010 Prepared by: New Jersey Department of Labor and Workforce Development, March 2010 Table 8 – Income Per Capita 53 Appendix 2 - Regression Results 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 Park City Price Equation Dependent Variable Usable Observations Degrees of Freedom Centered R2 Uncerentered R2 Mean of Dep. Variable Std. Error Dep. Variable Std. Error of Estimate Durbin Watson Statistic Variable Coeff Constant 162.44067 SQFT -0.05319 BD 7.33155 BA 5.74954 ESTIMATED 46.58589 CRSCTRDGE 29.49683 PRKAVE -13.36648 PDAY -15.52728 RSRTCTR 52.34432 SNWFLWR 75.05343 SNWCRST -19.48127 KNGS 0.00000 D82 -27.86515 D83 -17.98790 D84 -23.73294 D85 -46.40642 D86 -54.17574 D87 -61.82388 D88 -59.72268 D89 -50.88892 D90 -42.07275 D91 -33.48901 D92 -38.79587 D93 -28.94996 D94 -4.13900 D95 30.32111 D96 68.12208 D97 76.96135 D98 83.97963 D99 83.37392 D00 70.43984 D01 57.93130 D02 58.60940 D03 68.45588 D04 99.00437 D05 177.17003 D06 344.19221 D07 354.02536 D08 306.47834 D09 221.81667 D10 179.20066 D08 0.75420 D09 0.71390 D10 0.82722 Std Error 10.93705 0.00633 2.38084 2.75265 11.51448 5.35168 3.86007 5.19447 4.08142 3.92061 4.71674 0.00000 16.40575 14.28144 10.97678 10.82138 10.92280 11.01936 10.68071 10.23069 10.26986 11.02952 11.12932 10.42083 10.48495 10.90866 11.89957 11.93028 11.75968 11.93362 13.15115 11.81522 11.17499 10.64579 10.35815 10.60170 10.93143 12.73700 13.20239 13.17617 16.45350 0.06181 0.05725 0.06992 54 Linear PC PSQFT 1141 1101 0.9153 0.9752 184.3557 118.8623 35.1953 1.0524 T-Stat 14.85233 -8.40537 3.07940 2.08873 4.04585 5.51170 -3.46276 -2.98920 12.82503 19.14331 -4.13024 0.00000 -1.69850 -1.25953 -2.16210 -4.28840 -4.95988 -5.61048 -5.59164 -4.97414 -4.09672 -3.03631 -3.48592 -2.77809 -0.39476 2.77955 5.72475 6.45093 7.14132 6.98647 5.35617 4.90311 5.24469 6.43033 9.55812 16.71147 31.48647 27.79503 23.21385 16.83469 10.89134 12.20170 12.46939 11.83090 Signif 0.00000 0.00000 0.00213 0.03696 0.00006 0.00000 0.00056 0.00286 0.00000 0.00000 0.00004 0.00000 0.08970 0.20811 0.03083 0.00002 0.00000 0.00000 0.00000 0.00000 0.00004 0.00245 0.00051 0.00556 0.69310 0.00554 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 Park City Price Equation Exponential Dependent Variable LOG PC PSQFT Usable Observations 1141 Degrees of Freedom 1101 Centered R2 0.9317 Uncerentered R2 0.9991 Mean of Dep. Variable 5.0411 Std. Error Dep. Variable 0.5871 Std. Error of Estimate 0.1561 Durbin Watson Statistic Variable Coeff 1.2089 Std Error T-Stat Signif 1 Constant 5.05314 0.04853 104.13349 0.00000 2 SQFT -0.00035 0.00003 -12.31441 0.00000 3 BD 0.03947 0.01056 3.73686 0.00020 4 BA 0.07057 0.01221 5.77826 0.00000 5 ESTIMATED 0.06268 0.05109 1.22689 0.22013 6 CRSCTRDGE 0.06212 0.02374 2.61634 0.00901 7 PRKAVE -0.17330 0.01713 -10.11913 0.00000 8 PDAY -0.18157 0.02305 -7.87829 0.00000 9 RSRTCTR 0.31156 0.01811 17.20544 0.00000 10 SNWFLWR 0.38570 0.01740 22.17323 0.00000 11 SNWCRST -0.14921 0.02093 -7.13003 0.00000 12 KNGS 0.00000 0.00000 0.00000 0.00000 13 D82 -0.24817 0.07279 -3.40948 0.00067 14 D83 -0.22096 0.06336 -3.48721 0.00051 15 D84 -0.18286 0.04870 -3.75476 0.00018 16 D85 -0.35266 0.04801 -7.34524 0.00000 17 D86 -0.52647 0.04846 -10.86355 0.00000 18 D87 -0.58700 0.04889 -12.00631 0.00000 19 D88 -0.56096 0.04739 -11.83742 0.00000 20 D89 -0.44101 0.04539 -9.71577 0.00000 21 D90 -0.29965 0.04557 -6.57619 0.00000 22 D91 -0.25565 0.04894 -5.22412 0.00000 23 D92 -0.27062 0.04938 -5.48051 0.00000 24 D93 -0.21879 0.04624 -4.73201 0.00000 25 D94 -0.00837 0.04652 -0.17995 0.85722 26 D95 0.19150 0.04840 3.95657 0.00008 27 D96 0.41131 0.05280 7.79057 0.00000 28 D97 0.42614 0.05293 8.05063 0.00000 29 D98 0.49222 0.05218 9.43394 0.00000 30 D99 0.46636 0.05295 8.80807 0.00000 31 D00 0.41134 0.05835 7.04957 0.00000 32 D01 0.34381 0.05242 6.55861 0.00000 33 D02 0.36811 0.04958 7.42446 0.00000 34 D03 0.38998 0.04723 8.25643 0.00000 35 D04 0.51333 0.04596 11.16967 0.00000 36 D05 0.80409 0.04704 17.09451 0.00000 37 D06 1.23954 0.04850 25.55709 0.00000 38 D07 1.19406 0.05651 21.12942 0.00000 39 D08 1.08261 0.05858 18.48193 0.00000 40 D09 0.90411 0.05846 15.46533 0.00000 41 D10 0.85648 0.07300 11.73241 0.00000 42 D08 0.75420 0.06181 12.20170 0.00000 43 D09 0.71390 0.05725 12.46939 0.00000 44 D10 0.82722 0.06992 11.83090 0.00000 55 Deer Valley Price Equation Dependent Variable Usable Observations Degrees of Freedom Centered R2 Uncerentered R2 Mean of Dep. Variable Std. Error Dep. Variable Std. Error of Estimate Durbin Watson Statistic Variable Coeff Constant 195.41381 SQFT -0.04457 BD 6.33621 BA 6.54443 ESTIMATED -15.23419 ASPNWD -10.80041 CRCHVL 14.21784 DAYSTAR 16.78282 FAWNGRV 7.82101 LAKESIDE -2.98353 PINEINN 206.55328 PINNACLE 41.46703 PWDRRUN 67.99794 QNESTHER 15.67690 STNBRDGE 0.00000 D82 -8.21333 D83 -9.08186 D84 -25.44559 D85 -40.69757 D86 -43.51195 D87 -66.20709 D88 -56.25544 D89 -51.19309 D90 -44.24295 D91 -49.14128 D92 -48.09970 D93 -38.62359 D94 -29.76084 D95 -8.52166 D96 31.53628 D97 41.95352 D98 51.48673 D99 50.00029 D00 39.80783 D01 38.10879 D02 28.87965 D03 38.32283 D04 44.57063 D05 111.51562 D06 252.92460 D07 261.63890 D08 158.41566 D09 146.01905 D10 189.00081 Std Error 13.33080 0.00384 2.45821 2.47017 6.86106 5.24423 6.18651 6.24868 4.26003 4.62123 7.40357 5.41212 5.53112 4.89972 0.00000 13.96433 11.95503 12.03586 12.68962 12.30602 12.23543 11.95180 11.56919 11.86107 12.17619 11.77439 11.74625 11.70180 11.76263 12.02559 12.88809 12.68251 12.78804 12.24294 12.39202 12.62340 12.68363 11.70470 11.96286 12.59034 12.66697 14.69661 13.61268 16.62476 56 Linear DV PSQFT 957 914 0.8969 0.9760 167.4664 92.2087 30.2828 1.0887 T-Stat 14.65882 -11.60089 2.57757 2.64939 -2.22038 -2.05948 2.29820 2.68582 1.83590 -0.64561 27.89915 7.66188 12.29369 3.19955 0.00000 -0.58816 -0.75967 -2.11415 -3.20715 -3.53583 -5.41110 -4.70686 -4.42495 -3.73010 -4.03585 -4.08511 -3.28816 -2.54327 -0.72447 2.62243 3.25522 4.05966 3.90993 3.25149 3.07527 2.28779 3.02144 3.80793 9.32182 20.08878 20.65521 10.77906 10.72670 11.36864 Signif 0.00000 0.00000 0.01011 0.00820 0.02664 0.03973 0.02178 0.00737 0.06670 0.51869 0.00000 0.00000 0.00000 0.00142 0.00000 0.55657 0.44765 0.03477 0.00139 0.00043 0.00000 0.00000 0.00001 0.00020 0.00006 0.00005 0.00105 0.01115 0.46896 0.00888 0.00117 0.00005 0.00010 0.00119 0.00217 0.02238 0.00259 0.00015 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 Deer Valley Price Equation Exponential Dependent Variable LOG DVPSQFT Usable Observations 957 Degrees of Freedom 914 Centered R2 0.9264 Uncerentered R2 0.9994 Mean of Dep. Variable 5.0060 Std. Error Dep. Variable 0.4589 Std. Error of Estimate 0.1274 Durbin Watson Statistic Variable Coeff 1.4380 Std Error T-Stat Signif 1 Constant 5.22403 0.05607 93.17533 0.00000 2 SQFT -0.00027 0.00002 -16.54166 0.00000 3 BD 0.05773 0.01034 5.58428 0.00000 4 BA 0.01962 0.01039 1.88830 0.05930 5 ESTIMATED -0.04424 0.02886 -1.53295 0.12563 6 ASPNWD -0.03162 0.02206 -1.43352 0.15205 7 CRCHVL 0.12598 0.02602 4.84179 0.00000 8 DAYSTAR 0.15183 0.02628 5.77710 0.00000 9 FAWNGRV 0.06335 0.01792 3.53567 0.00043 10 LAKESIDE 0.04413 0.01944 2.27055 0.02341 11 PINEINN 0.86995 0.03114 27.93875 0.00000 12 PINNACLE 0.26716 0.02276 11.73689 0.00000 13 PWDRRUN 0.40730 0.02326 17.50869 0.00000 14 QNESTHER 0.11685 0.02061 5.67018 0.00000 15 STNBRDGE 0.00000 0.00000 0.00000 0.00000 16 D82 -0.03479 0.05873 -0.59244 0.55370 17 D83 -0.08498 0.05028 -1.69019 0.09133 18 D84 -0.15274 0.05062 -3.01739 0.00262 19 D85 -0.22389 0.05337 -4.19500 0.00003 20 D86 -0.26932 0.05176 -5.20368 0.00000 21 D87 -0.38921 0.05146 -7.56334 0.00000 22 D88 -0.46099 0.05027 -9.17084 0.00000 23 D89 -0.45130 0.04866 -9.27509 0.00000 24 D90 -0.31891 0.04989 -6.39297 0.00000 25 D91 -0.35298 0.05121 -6.89274 0.00000 26 D92 -0.34557 0.04952 -6.97829 0.00000 27 D93 -0.27609 0.04940 -5.58869 0.00000 28 D94 -0.20449 0.04922 -4.15505 0.00004 29 D95 -0.01813 0.04947 -0.36652 0.71406 30 D96 0.20781 0.05058 4.10870 0.00004 31 D97 0.25947 0.05420 4.78693 0.00000 32 D98 0.31112 0.05334 5.83274 0.00000 33 D99 0.27747 0.05378 5.15890 0.00000 34 D00 0.25835 0.05149 5.01731 0.00000 35 D01 0.23682 0.05212 4.54385 0.00001 36 D02 0.19843 0.05309 3.73746 0.00020 37 D03 0.21836 0.05334 4.09345 0.00005 38 D04 0.27188 0.04923 5.52289 0.00000 39 D05 0.53963 0.05031 10.72533 0.00000 40 D06 0.97806 0.05295 18.47056 0.00000 41 D07 1.01052 0.05327 18.96806 0.00000 42 D08 0.75420 0.06181 12.20170 0.00000 43 D09 0.71390 0.05725 12.46939 0.00000 44 D10 0.82722 0.06992 11.83090 0.00000 57 The Canyons Price Equation Linear Dependent Variable CN PSQFT Usable Observations 896 Degrees of Freedom 860 Centered R2 0.9378 Uncerentered R2 0.9820 Mean of Dep. Variable 126.2498 Std. Error Dep. Variable 80.5453 Std. Error of Estimate 20.4868 Durbin Watson Statistic Variable 1 Constant Coeff 1.3100 Std Error T-Stat Signif 122.88172 4.62732 26.55571 0.00000 2 SQFT -0.01183 0.00177 -6.68959 0.00000 3 BD 1.58176 1.64112 0.96383 0.33540 4 BA -6.74591 1.43887 -4.68835 0.00000 5 ESTIMATED -4.15288 5.22088 -0.79544 0.42658 6 PKWV -16.70125 3.94314 -4.23552 0.00003 7 PKWHC -12.53119 1.73209 -7.23471 0.00000 8 REDPINE 0.00000 0.00000 0.00000 0.00000 9 D82 5.76702 6.02813 0.95669 0.33899 10 D83 -15.07629 5.31365 -2.83728 0.00466 11 D84 -19.96132 5.98171 -3.33706 0.00088 12 D85 -23.09929 6.58298 -3.50894 0.00047 13 D86 -44.79446 5.95452 -7.52276 0.00000 14 D87 -42.02295 5.61575 -7.48305 0.00000 15 D88 -40.31275 5.43067 -7.42317 0.00000 16 D89 -35.45655 5.31125 -6.67575 0.00000 17 D90 -30.63194 5.61457 -5.45580 0.00000 18 D91 -25.56679 5.80190 -4.40662 0.00001 19 D92 -23.03146 5.35395 -4.30177 0.00002 20 D93 -16.70496 5.16065 -3.23699 0.00125 21 D94 9.77136 5.47384 1.78510 0.07460 22 D95 34.25297 5.58158 6.13679 0.00000 23 D96 56.35550 5.82779 9.67013 0.00000 24 D97 63.13823 5.35913 11.78144 0.00000 25 D98 79.96018 5.81219 13.75733 0.00000 26 D99 72.55646 6.66830 10.88080 0.00000 27 D00 62.61223 6.42367 9.74711 0.00000 28 D01 67.03060 6.65022 10.07945 0.00000 29 D02 62.31909 6.63641 9.39048 0.00000 30 D03 52.45061 5.76022 9.10565 0.00000 31 D04 73.79207 5.42560 13.60073 0.00000 32 D05 153.29625 5.25736 29.15840 0.00000 33 D06 248.47321 5.76552 43.09640 0.00000 34 D07 261.25604 6.87050 38.02576 0.00000 35 D08 190.19640 7.60626 25.00524 0.00000 36 D09 116.86278 7.38219 15.83037 0.00000 37 D10 105.57185 15.14023 6.97294 0.00000 38 D07 1.19406 0.05651 21.12942 0.00000 39 D08 1.08261 0.05858 18.48193 0.00000 40 D09 0.90411 0.05846 15.46533 0.00000 41 D10 0.85648 0.07300 11.73241 0.00000 42 D08 0.75420 0.06181 12.20170 0.00000 43 D09 0.71390 0.05725 12.46939 0.00000 44 D10 0.82722 0.06992 11.83090 0.00000 58 The Canyons Price Equation 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 Dependent Variable Usable Observations Degrees of Freedom Centered R2 Uncerentered R2 Mean of Dep. Variable Std. Error Dep. Variable Std. Error of Estimate Durbin Watson Statistic Variable Coeff Constant 4.78948 SQFT -0.00021 BD 0.01640 BA -0.00390 ESTIMATED -0.03108 PKWV -0.11504 PKWHC -0.13049 REDPINE 0.00000 D82 0.02587 D83 -0.18910 D84 -0.25676 D85 -0.32117 D86 -0.63750 D87 -0.57516 D88 -0.55849 D89 -0.53160 D90 -0.39085 D91 -0.31264 D92 -0.27783 D93 -0.17833 D94 0.08349 D95 0.30171 D96 0.46149 D97 0.49546 D98 0.61249 D99 0.55517 D00 0.49850 D01 0.53383 D02 0.49615 D03 0.44855 D04 0.56877 D05 0.93544 D06 1.29609 D07 1.31024 D08 1.11569 D09 0.83025 D10 0.78287 D07 1.19406 D08 1.08261 D09 0.90411 D10 0.85648 D08 0.75420 D09 0.71390 D10 0.82722 Exponential Std Error 0.02952 0.00001 0.01047 0.00918 0.03330 0.02515 0.01105 0.00000 0.03845 0.03389 0.03816 0.04199 0.03798 0.03582 0.03464 0.03388 0.03581 0.03701 0.03415 0.03292 0.03492 0.03560 0.03717 0.03418 0.03707 0.04254 0.04098 0.04242 0.04233 0.03674 0.03461 0.03354 0.03678 0.04383 0.04852 0.04709 0.09658 0.05651 0.05858 0.05846 0.07300 0.06181 0.05725 0.06992 59 LOGCNPSF 896 860 0.9519 0.9993 4.6625 0.5839 0.1307 1.4071 T-Stat 162.26410 -18.28761 1.56646 -0.42482 -0.93314 -4.57369 -11.81088 0.00000 0.67290 -5.57909 -6.72910 -7.64851 -16.78405 -16.05612 -16.12214 -15.69095 -10.91325 -8.44763 -8.13522 -5.41745 2.39105 8.47413 12.41421 14.49358 16.52056 13.05195 12.16587 12.58426 11.72051 12.20766 16.43442 27.89395 35.24196 29.89686 22.99519 17.63148 8.10622 21.12942 18.48193 15.46533 11.73241 12.20170 12.46939 11.83090 Signif 0.00000 0.00000 0.11761 0.67107 0.35101 0.00001 0.00000 0.00000 0.50119 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.01701 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 0.00000 Price Index – Log v. Linear 300 250 200 150 RlPrice(log) Real Price 100 50 Figure 16 – PC Real Price Index – Linear vs. Log 60 2010 2008 2006 2004 2002 2000 1998 1996 1994 1992 1990 1988 1986 1984 1982 1980 0 Park City Ski Area Ski Day Equation Regression Statistics Multiple R 0.97701 R Square 0.954549 Adjusted R Square 0.948621 Standard Error 79837.28 Observations 27 ANOVA df Regression Residual Total SS MS F Significance F 3 3.08E+12 1.03E+12 161.0135 1.41E-15 23 1.47E+11 6.37E+09 26 3.23E+12 Coefficients Standard Error t Stat Intercept (PCSkidays)-1099974 267651.2 -4.10973 Sdays t-1 0.414106 0.135763 3.050222 PCSnFall 563.2646 178.2133 3.160621 U.S. Inc t-1 124.5875 30.63429 4.066928 P-value Lower 95%Upper 95%Lower 95.0% Upper 95.0% 0.000428 -1653653 -546296 -1653653 -546296 0.005678 0.13326 0.694952 0.13326 0.694952 0.004371 194.6024 931.9268 194.6024 931.9268 0.000476 61.2156 187.9593 61.2156 187.9593 61 Park City Construction Permits Equation - Park City Prices SUMMARY OUTPUT Regression Statistics Multiple R 0.577994 R Square 0.334077 Adjusted R Square 0.254166 Standard Error 125.215 Observations 29 ANOVA df Regression Residual Total SS MS F Significance F 3 196641.4 65547.12 4.180621 0.015758 25 391970 15678.8 28 588611.3 Coefficients Standard Error t Stat Intercept (permits) 197.5641 90.30908 2.187643 Prmt t-1 -0.06641 0.17555 -0.37828 Stock t-1 -0.04872 0.01659 -2.93648 Real Price 2.273626 0.729426 3.117008 P-value Lower 95%Upper 95%Lower 95.0% Upper 95.0% 0.038259 11.56903 383.5591 11.56903 383.5591 0.708413 -0.42796 0.295144 -0.42796 0.295144 0.00703 -0.08289 -0.01455 -0.08289 -0.01455 0.004551 0.771345 3.775907 0.771345 3.775907 Park City Permits Equation - Deer Valley Prices SUMMARY OUTPUT Regression Statistics Multiple R 0.539582 R Square 0.291149 Adjusted R Square 0.206086 Standard Error 129.1879 Observations 29 ANOVA df Regression Residual Total Intercept Prmt t-1 Stock t-1 Real Price SS MS F Significance F 3 171373.3 57124.45 3.422774 0.032554 25 417238 16689.52 28 588611.3 Coefficients Standard Error t Stat 45.08003 124.2797 0.362731 -0.04375 0.180315 -0.24265 -0.02261 0.013859 -1.63178 2.316019 0.839372 2.759229 P-value Lower 95%Upper 95%Lower 95.0% Upper 95.0% 0.719855 -210.879 301.0388 -210.879 301.0388 0.810258 -0.41512 0.327612 -0.41512 0.327612 0.11526 -0.05116 0.005928 -0.05116 0.005928 0.010682 0.5873 4.044737 0.5873 4.044737 62 Park City Price Time Series Regression SUMMARY OUTPUT Regression Statistics Multiple R 0.92940017 R Square 0.86378467 Adjusted R Square 0.84675776 Standard Error 16.1174596 Observations 28 ANOVA df Regression Residual Total Intercept (Rprice) Price T-1 PC SkiDays Stock t-1 SS MS F Significance F 3 39535.2004 13178.4001 50.7305426 1.5374E-10 24 6234.54012 259.772505 27 45769.7405 CoefficientsStandard Error t Stat 5.65263577 10.9270373 0.51730726 0.49724341 0.13035453 3.8145464 0.00011644 3.395E-05 3.42989046 -0.0144561 0.00595067 -2.4293179 P-value 0.60967696 0.00084077 0.00219043 0.02298241 Lower 95% -16.899661 0.22820488 4.6375E-05 -0.0267377 Upper 95% Lower 95.0% Upper 95.0% 28.2049322 -16.899661 28.20493215 0.76628194 0.22820488 0.766281941 0.00018651 4.6375E-05 0.000186512 -0.0021745 -0.0267377 -0.00217449 Deer Valley Price Time Series Regression SUMMARY OUTPUT Regression Statistics Multiple R 0.861852554 R Square 0.742789825 Adjusted R Square 0.710638554 Standard Error 14.85530371 Observations 28 ANOVA df Regression Residual Total Intercept Price T-1 Skier Days Stock t-2 SS MS F Significance F 3 15295.093 5098.3642 23.102969 2.95E-07 24 5296.3212 220.68005 27 20591.414 CoefficientsStandard Error t Stat 27.27227425 14.932799 1.8263337 0.51011256 0.1346865 3.7874063 7.55996E-05 3.079E-05 2.4554419 -0.01033031 0.0058615 -1.7623998 P-value 0.0802662 0.0009 0.0216959 0.0907347 63 Lower 95% Upper 95% Lower 95.0%Upper 95.0% -3.5475086 58.092057 -3.5475086 58.0920571 0.2321333 0.7880919 0.2321333 0.78809186 1.206E-05 0.0001391 1.206E-05 0.00013914 -0.0224279 0.0017672 -0.0224279 0.00176724 Bibliography (NSAA) National Ski Areas Association. "Estimated U.S. Ski Industry Visits by Region 1978/79 - 2008/09." 2009. www.nsaa.org. 1 6 2010 <http://www.nsaa.org/nsaa/press/historicalvisits.pdf>. DiPasquale, Denise and William C. Wheaton. Urban Economics and Real Estate Markets. Prentice-Hall, Inc. , 1996. IBIS World. "Ski Resorts in the US, IBIS World Industry Report 71392." January 2010. IBIS World. 6 June 2010 <http://www.ibisworld.com/industryus/default.aspx?indid=1653>. Lee, Sean. "Second Home Real Estate Market: Economic Analysis of Residential Pricing Behavior Near Heavenly Ski Resort, CA." 2008. Miller, Norman G. "Residential Property Hedonic Pricing Models: A Review." Research in Real Estate, Vol. 2. JAI Press Inc., 1982. 31-56. National Association of Realtors. Second Homes: Talking Points. 10 March 2010. 6 July 2010 <http://www.realtor.org/press_room_secured/public_affairs/tpsecondhomes>. Park City Chamber and Visitor's Bureau. "Economic and Relocation Package - Park City History." 2010. ParkCityInfo.com. 5 June 2010 <http://www.parkcityinfo.com/docs/PARK_CITY%20HISTORY%202009.pdf>. Park City Municipality. "Park City: Quick Facts." 1 1 2010. ParkCity.org. 6 6 2010 <http://www.parkcity.org/index.aspx?page=279>. Ski Utah. "Utah Skier Days Table." 24 6 2010. www.skiutah.org. 24 6 2010 <http://www.skiutah.com/media/story_starters/utah-skier-days-table>. Wheaton, William C. "Resort Real Estate: Does Supply Prevent Appreciation?" Journal of Real Estate Research, Vol.27 27 (2005): 1-16. 64